fix: group_by multiple null columns produce phantom row (#15659)

pola-rs · Apr 15, 2024 · f1e231e · f1e231e
1 parent 2a9a217
commit f1e231e
Show file tree

Hide file tree

Showing 2 changed files with 11 additions and 1 deletion.
diff --git a/crates/polars-core/src/frame/group_by/mod.rs b/crates/polars-core/src/frame/group_by/mod.rs
@@ -76,8 +76,13 @@ impl DataFrame {
                 .cloned()
                 .collect::<Vec<_>>();
             if by.is_empty() {
+                let groups = if self.height() == 0 {
+                    vec![]
+                } else {
+                    vec![[0, self.height() as IdxSize]]
+                };
                 Ok(GroupsProxy::Slice {
-                    groups: vec![[0, self.height() as IdxSize]],
+                    groups,
                     rolling: false,
                 })
             } else {

diff --git a/py-polars/tests/unit/operations/test_group_by.py b/py-polars/tests/unit/operations/test_group_by.py
@@ -985,3 +985,8 @@ def test_aggregated_scalar_elementwise_15602() -> None:
     )
     expected = pl.DataFrame({"group": [1, 2], "foo": [[True, True], [True]]})
     assert_frame_equal(out, expected)
+
+
+def test_group_by_multiple_null_cols_15623() -> None:
+    df = pl.DataFrame(schema={"a": pl.Null, "b": pl.Null}).group_by(pl.all()).len()
+    assert df.is_empty()